People join the program right after a bad year

Question

An analyst evaluates a job-training program with difference-in-differences, comparing trainees' earnings change to a control group of non-trainees. The estimate is strongly positive. But you notice that people tend to enroll right after a year in which their earnings dropped sharply.

Explain why enrolling after a bad year can create a fake positive effect, and how it breaks parallel trends.

Accepted Answer

Enrollment is triggered by a temporary dip. People often sign up for training just after losing a job or suffering an unusually bad income year. That low point is frequently temporary: earnings tend to bounce back toward their normal level on their own, a pattern called regression to the mean. So trainees are measured starting from an artificially low base, and their natural recovery makes the program look effective even if it does nothing. This pre-enrollment dip and rebound is sometimes called Ashenfelter's dip. Why parallel trends is violated. The control group of non-trainees did not just…

People join the program right after a bad year

Your answer

More Statistics questions